Development of Machine Learning-Driven Dual-Task Gait Test Model for Cognitive Impairment Screening.

Area of Science:

• Neurology
• Gerontology
• Artificial Intelligence

Background:

• Cognitive impairment, including dementia and mild cognitive impairment, affects a significant portion of the aging population.
• Early detection of cognitive decline is crucial for timely intervention and management.
• Current screening methods can be resource-intensive and may not be suitable for high-throughput application.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI)-aided dual-task gait test model.
• To enable scalable, high-throughput screening for cognitive impairment in older adults.
• To assess the diagnostic performance of AI-assisted gait analysis in identifying dementia and mild cognitive impairment.

Main Methods:

• A case-control study involved 201 community-dwelling adults aged ≥60 years, categorized into dementia, mild cognitive impairment, and cognitively intact groups.
• Participants underwent single-task and dual-task gait assessments (serial subtraction, animal naming).
• AI-assisted gait analysis extracted 45 gait parameters; logistic regression and machine learning models evaluated diagnostic accuracy (AUC).

Main Results:

• Gait speed and stride variability were key predictors of cognitive status.
• Logistic regression models showed high screening accuracy for dementia (AUC=0.883) and cognitive impairment (AUC=0.895).
• Machine learning models, particularly support vector machines, further validated these findings with AUCs of 0.786 (dementia) and 0.808 (cognitive impairment).

Conclusions:

• The AI-aided dual-task gait paradigm offers a scalable, cost-effective, and accurate method for population-level cognitive impairment screening.
• This approach facilitates early detection of cognitive decline in community settings.
• It holds transformative potential for proactive dementia prevention strategies through timely interventions.